On Fri, 3 Jul 2026 15:54:34 GMT, Alan Bateman <[email protected]> wrote:
> JVMTI GetLocalObject/GetLocalInstance: Right now, the behavior when preview > features is enabled is buried in the description of the value_ptr parameter. > Add text to the function descriptions so that it is more discoverable. > > JDWP: I realize there has been some flip-flop on this but I think we need the > "When preview features are enabled" text and remove any possibility of > developers not understanding that the feature is in preview. > > JDI: The first use of "value object" or "value class" in the updated methods > now link to Class::isValue. This will ensure that the "PREVIEW" superscript > appears in the usage. It might be that we come up with a better way to > communicate that the behavior is for value objects which are in preview for > this release. > > JDI StackFrame.getValue/getValues: JDI has always filtered out "this" from > the local variables so I think the only way for JDI tools to get a local > variable for an object under construction is if the constructor invokes a > method with "this" as a parameter (this-escape). Instance fields in value > objects are strictly-initialized final fields so any leaking of "this" will > be after the object has been initialized. I think we can drop the text added > to these two methods. If this get a "this" for a value object under > construction then it will already be initialized, and setValue will fail as > the fields are final. We need to keep the text in StackFrame.thisObject, > which is tweaked slightly to make it a bit clearer. Future work might be > needed here to have ObjectReference.getValue fail if the field has not been > initialized. > > --------- > - [x] I confirm that I make this contribution in accordance with the [OpenJDK > Interim AI Policy](https://openjdk.org/legal/ai). src/jdk.jdi/share/classes/com/sun/jdi/ThreadReference.java line 460: > 458: * a {@linkplain Class#isValue() value class}. Forcing early return > 459: * on a thread with only one frame on the stack causes the thread to > 460: * exit when resumed. Do you want this update in the JDWP spec also? ------------- PR Review Comment: https://git.openjdk.org/valhalla/pull/2627#discussion_r3521318154
